University of Venice

University of Venice Working Papers in Linguistics

Vol. 21, 2011

University of Venice

1. Away with N

-of-N

constructions!

Much (if not all) of the literature written on pseudopartitives so far has dealt with the so- called N-of-N constructions, or, in another denomination, the N

-of-N

constructions. But nothing has been said about constructions in which, for some more or less strange reason, N

happens to be the same as N

. The obvious question is why… Is there no such construction as the N

-of-N

construction? Or is it simply the case that linguists have not paid enough attention to the empirical matter they were supposed to deal with? The aim of the following paper is to try and give an answer to the questions above. In so doing, the paper will mainly focus on Romanian data, and test whether constructions such as băiat de băiat (‘boy of boy’), fată de fată (‘girl of girl’), sentiment de sentiment (‘feeling of feeling’) a. o. can be subsumed under the name of “qualitative N-of-N constructions”.

2. N-of-N constructions

2.1. Some general ideas concerning ‘N-of-N’ constructions

But, before delving into the rather confusing realm of N

-de-N

constructions, let us first

present some general ideas concerning N-of-N constructions. According to the traditional

classification (cf. Selkirk (1977)), constructions which include a noun followed by of, and

then again by a noun, can be divided into two major classes:

(i) partitives

(1) a group of the students

in which case N

is preceded by the definite article the (indicating a specific referent), and (ii) pseudopartitives

in which case N

is not preceded by a definite article. “With partitive constructions, N

denotes a definite or delimited domain, while with pseudopartitive constructions, N

refers to an indefinite or unrestricted domain” (Dogaru (2009): 81).

Pseudopartitives in their turn can be divided into two classes:

(a) quantitative pseudopartitives (which Corver (1998) simply refers to as

“pseudopartitives”):

(2) a cup of coffee and

(b) qualitative pseudopartitives (which Corver (1998) refers to by the term “N-of-N constructions”, or “binominal”- we will use the term “binominal” in this sense throughout the paper):

(3) a beauty of a woman

At first sight, we could say that, in all the constructions above, we have an N

followed by “of”, followed by N

. However, the lexical status of N

has been highly disputed in the literature, and linguists such as Selkirk (1977) or Dogaru (2009) have argued that it is rather functional or semilexical (for it is not really an N

).

If we make it our purpose to detect those constructions in which it is possible for N

to

be the same as N

, what we can easily notice is that, out of the constructions present above,

the only one for which such an identity would be conceivable in principle is the qualitative

pseudopartitive construction. Of course, we can also say:

(4) a student of the students

but, by student, we do not understand a part of the class of students , but a noun designating a collection of students, belonging to the bigger class. We could argue that, actually, such constructions are to be analyzed as partitives, which would be supported by the fact that, in English, we have more or less expressions such as king of kings, sun of suns. However, we argue that it is best not to analyze them as partitives, given (a) the absence of the definite article in the case of the N

and (b) the different meaning (superlative meaning). As for quantitative pseudopartitive constructions, they do not allow identity between N

and N

. In (5):

(5) *a coffee of coffee

we cannot speak of identity between the two nouns. Not only does such an expression sound odd, but if we were to make sense of it, what we would get would be something with the approximate meaning of “a cup of coffee”, in which case it is again clear that the two nouns are differentiated by means of the mass/ count distinction: “coffee” means “a unit of coffee” (in this case, “a cup”), while the second noun “coffee” refers to the substance. We cannot really speak of identity. Not only does N

have a different meaning from N

, but the difference between them is also reflected in the syntax (the presence of the indefinite article a vs. the absence of the indefinite article in the case of the second noun). We base our reasoning on the presence of a silent noun in (7):

(6) “Oh, woman, give me a coffee, for God’s sake! Can’t you see I’m tired?”

(7) “Oh, woman, give me a CUP (of) coffee, for God’s sake! Can’t you see I’m tired?”

A coffee of coffee is an odd and unnecessary expression. If we use “coffee” to refer to the

recipient of coffee, then there is no need to indicate the substance contained in the recipient

(for it is already mentioned). The ungrammaticality of such an expression could very well

serve as support for a silent noun analysis of metonymic nouns. An expression such as *a

UNIT/ CUP of coffee of coffee would be redundant: there is no need to further add “of

coffee” and, thus, a coffee of coffee is neatly ruled out (since it is already there). Thus, in

the case of partitives and quantitative pseudopartitives, we simply cannot speak of N

-of-N

constructions. However, this is not how things stand in the case of qualitative pseudopartitives, or binominal constructions. Or, at least, we are in the realm of conceivability. Why is it then that we simply have not heard expressions such as:

(8) a beauty of a beauty or (9) an oaf of an oaf?

How come our ears have not detected such sounds? Why is it that, if we resort to a very simple search on google, no such expressions will be found? What is the mysterious reason that lies hidden behind this puzzling absence?

2.2. A beauty of a beauty…

Let us simply prick our mind’s ears and get hold of a possible explanation. If I am walking through the park one day and I see a most incredibly looking dog, called Anita, when I go home, obviously still thinking of Anita’s gorgeous looks, I will tell my mother:

(10) I saw a beauty of a dog.

Could I not instead say something like:

(11) ??I saw a beauty of a beauty. ?

If not, why is that? Why is it that (10) is perfectly fine, while (11) sounds odd? The reason

for this is fairly simple, we will venture to say. As we very well know, a great deal in the

literature on binominals has been written on the type of nouns which may occupy the N

or

the N

position. Milner (1978) has argued that N

has to be an evaluative noun, it has to

express the speaker’s evaluation of a particular person/animal/object (entity). Other

linguists (Matushansky (2002), Vişan (2003)) have spoken about the fact that the noun has

to be scalar in interpretation. Irrespective of the formulation, the idea is basically the same:

not any noun can occupy the position of N

, but only those nouns which express the speaker’s subjective view upon the entity denoted by N

. The class of N

nouns is limited (a fact which has been used as an argument in favour of the semi-lexical nature of N

). So far, so good. But what about the class of nouns in N

? Is there any kind of restriction in the case of N

? As we can easily see, (10) is perfectly fine. N

denotes a dog, evaluated by the speaker through the words “a beauty”: in other words, I saw a dog and the dog was a beauty. In (11), things are not that simple: what I am saying is that I saw a beauty and that the beauty (I saw) was a beauty. An important remark is in order here. Please note that I am not actually saying what I saw (a dog, a cat, a man, a woman or a flower). I am already referring to the dog by means of an evaluative noun. Hence, when I want to evaluate the entity/ animal I saw I will evaluate it once more. “A beauty” in the position of N

acts as a referential noun, “a beauty” in the position of N

acts as an evaluation of the referent of N

. Now, we have to admit that this is fairly odd. Unless I want to make it clear that I am dealing with a beauty, there is no reason for such redundancy.

In other words, even if it seems evaluative, N

picks up the referent (by means of a certain trait), and N

evaluates N

with respect to another trait:

(12)

Types of N-of- N constructions

Positive- Positive

Positive- Negative

Negative- Positive

Negative- Negative Examples a wonder of a

beauty

a cutie of a jerk

that jerk of a genius

?an asshole of a jerk

What we mean when we refer, for example, to a handsome man by the construction a beauty of a beast is not that he is a beast and he is a beauty (the relation is not one of coordination), but we pick him out as a beast (from the set of beasts), and we say that this beast (actually, man) we are referring to is one to which the property of being a beauty can be ascribed (That beauty is a beast).

This is why it would be very strange to pick out an asshole and say that the property of being an asshole can be ascribed to him. It obviously can, since he is a member of the set of assholes. There is no need for that. Further

1

. In other words, N

is extensional, it picks an element out of a class, while N

is intensional (it ascribes a

property to the element selected by N

).

specifications are required to describe the asshole, not redundancies. Moreover, this is also the reason why, when N

and N

are near synonyms (an asshole of a jerk), the construction sounds odd (jerks are obviously assholes).

In conclusion, N

can be argued to be referential, and N

to be evaluative.

When I am saying:

(13) Tibi kissed a withered leaf of a woman.

what I mean is:

(14) Tibi kissed a woman. and not (15) #Tibi kissed a withered leaf.

Some might argue that, in the examples above, N

also has a concrete denotation,

and, hence, it serves our purposes far too well, because (15) is implausible. For those who might argue thus, let us take an example such as:

(16) Jane dated an oaf of a smartass.

In this example, oaf does not have any concrete meanings. But, despite this, (16) still means that:

2

. In a sense, we always do refer to entities by means of nouns which express our perception of them. The only difference is that when we are dealing with kind nouns, our perception coincides with the others’

perception (we all agree that Anita is a dog), whereas when we are dealing with evaluative nouns, our perception may be different (we might deem Anita a beauty, whereas some might totally resent fluffy white fur, and consider Anita a caricature of a dog.)

3

. Another funny example translated here (from Dutch) given by Corver (1998) is that, fortunately, Jan heft [een droom van een huis] gekocht, i.e. “Jan bought a dream of a house”, does not mean “Jan bought a dream”, but it means “Jan bought a house.”

4

. In Dogaru’s terms, it is a noun which is coerced into an evaluative reading, not one which is evaluative by

nature.

(17) Jane dated a smartass. and not

(18) #Jane dated an oaf. (although (16) is perfectly OK from a grammatical standpoint).

Of course, in a sense, it does, because the smartass Jane dated was an oaf, but this is a conclusion the interlocutor reaches (through reasoning), not an assertion the speaker makes:

A: Jane dated a smartass. B: The smartass she dated was an oaf. Hence, C: Jane dated an oaf.

To conclude, the reason why N

cannot be identical to N

would be redundancy. As already mentioned above, since the relation between N

and N

is not one of coordination, but one of predication (N

is predicated of N

), it is fairly odd to say something like an idiot of an idiot, i.e. to pick an idiot out of the set of idiots and then argue that the idiot you have chosen to refer to is such that the property of being an idiot can be predicated of him. It obviously can.

And so, this is the reason why N

and N

in English binominals are not identical. Speakers choose different nouns for a very simple reason: reason itself. However, this explanation has its problems given the fact that, if we assume that N

in “a BEAUTY of a beauty” has a different meaning from N

, namely, the meaning ‘true/ authentic beauty’, there is no redundancy. The idea of a different meaning would be supported by the existence of expressions such as:

5

. This is something we should keep in mind, because it gives us a totally different perspective upon what has been generally labelled in the literature as “the semantic transparency” of pseudopartitives. (Corver (1998), Den Dikken (1998)). According to them, in an example such as:

(i) Billy drank a cold glass of beer.

the adjective cold refers both to the glass and the beer. We would like to suggest that (i) actually expresses the idea that Bill drank cold beer. But, because the beer was in the glass, the glass became cold as well. What the speaker says is not that Billy drank a cold glass. The idea of the coldness of the glass is reached through world knowledge and reasoning: A: Billy drank cold beer. B: The beer was in a glass. Hence C: The glass of beer was cold as well (because, generally, the container acquires the temperature of the liquid therein- this is world knowledge.).

6

. The idiot is an idiot is an analytic sentence, true, irrespective of the meaning of the word idiot. The

sentence is true by virtue of its form.

(19) a man’s man (20) a doctor’s doctor

We will later on explore this idea, showing that, in fact, we are not dealing with two different nouns, but with the same noun inserted in different places in the derivation.

3. In a Romanion Fashion……

We have seen how things stand in English in the case of N

-of-N

constructions. Now, let’s delve into the Romanian data, and see what it is that we get. Our focus will be on constructions which have initially entered the language as substandard, but which have become very fashionable in contemporary Romanian, not so fashionable as to become standard, but, nevertheless, recurrent in spoken Romanian: băiat de băiat (‘boy of boy’), fată de fată (‘girl of girl’), femeie de femeie (‘woman of woman’), sentiment de sentiment (‘feeling of feeling’) etc. If, initially, it was băiat de băiat that stealthily crept into the language, the pattern soon generalized itself, generating expressions where any type of noun can take the place of N, such as iepure de iepure (‘bunny of bunny’), gogoasă de gogoaşă (‘donut of donut’), trandafir de trandafir (‘rose of rose’), maşină de maşină (‘car of car’), etc.

What we will like to test in what follows is whether this type of construction, which proves to be such a productive pattern in Romanian, is a binominal construction, a qualitative pseudopartitive construction, or not. Our intuition is that it is not. We would like to claim that, despite its deceiving appearance, băiat de băiat does not in fact count as a pseudopartitive.

3.1. Speculation-the Mother of Creation

We do not know which the first expression was, but we will speculate that it was băiat de băiat (‘boy of boy’), by far the most popular of all, as in the lines “(Ia uitaţi-vă la mine./

Sunt bărbat şi îmi stă bine. )/ Mă numesc băiat băiat./…../ Că sunt băiat de băiat…”

(Nicolae Guţă, Băiat de băiat), ‘(Look at me,/ I am a man and I feel good about it.)’, the last two lines literally translate as ‘I am boy boy …/Because I am boy of boy.’. The meaning of this construction is “superbăiat”, “an awesome boy/ guy”, but what is understood by “awesome” is here very different from the ideal use of this word. To be

“băiat de băiat” (boy-of-boy) means to have money and girls, to have a rich dad who can fulfill all your wishes, to make use of people so as to serve your purposes, and to be very proud of it, what we would say in colloquial language- “to be a smartass”.

In relation to the origin of this expression, several possible hypotheses come to mind.

(a) On a first hypothesis, to be “băiat de băiat” means coming from a rich, but rather uneducated family, it says something about the origin of the boy we are speaking about. It may be similar with the pattern băiat de doctor (‘boy of doctor’), băiat de avocat (‘boy of lawyer’), băiat de deputat (‘boy of deputy’) a.o. The second occurrence of the noun băiat does not refer to the same individual, but to the (social, moral) status of that individual’s father, to his family roots. Thus, băiat de băiat would in this sense be more or less similar to băiat din băiat, băiat care se trage din băiat (‘boy coming from a boy’). This would go very much in line with the possibility existent in Romanian to use de (‘of’) instead of din (‘from’) in various contexts

(băiat de doctor, fată de medic), and it could be subsumed under the label of “analytic Genitive” (the expression of the Genitive case in an analytic fashion, by means of prepositions (as in the case of la mijloc de codru ‘at middle of forest’, rather than in a synthetic fashion, by means of markers attached to the end of the word:

băiatul doctorului (‘boy-the doctor-GEN’), fată medicului (‘girl-the doctor-GEN’).

Moreover, the substitution of din by de is also supported by facts from the history of language (where we know that din is a compound form from de and în).

(b) On a second hypothesis,băiat de băiat can be understood as băiat din băiat, but what we understand by means of the second noun băiat is not our boy’s father, but rather an ideal prototypical boy. Băiat de băiat would mean ‘băiat care face cinste numelui de “băiat”’, ‘un superbăiat’, ‘un băiat de calitate’, that is, ‘a boy worthy of the name ‘boy’’, ‘a cool boy’. In this case, the ideal prototype is a caricature, because everything has an ironic flavour,

7

. As Dogaru (2009) points out, although the partitive can be expressed by means of de, Romanian makes

use of two typical prepositions (din, dintre): zece grame de/ din brânză. (‘ten grams of/ from (meaning ‘of

the’) cheese’).

nothing is that ideal. It is as if there existed an ideal world where each entity in the real world had a sort of prototype, endowed with all the specific qualities of that entity. In this case, it is the ideal world for those who dream of money, cars:

(21) băiat

de băiat

Although, apparently, N

and N

represent the same noun, their meaning is slightly different: the first noun băiat is used in a normal fashion, meaning ‘young male’, whereas the second noun băiat refers to more than that, ‘young male, with lots of money, girls and cars…’. If we say:

(22) Gigi e băiat de băiat.

‘Gigi is boy of boy.’

this means Gigi is a boy/ guy belonging to the class of cool boys/ guys. He is not any sort of guy. The PP de băiat modifies the noun băiat in an essential way. The hidden message would more or less sound this way: there are many băieţi in this world, but few băieţi de băieţi, or, in other words, many boys, few real boys, and Gigi is one of them.

Summing up the remarks above, what we can say is that, from a purely intuitive, interpretative point of view, what we get is a structure in which the first noun băiat is felt to be the head, and the PP de băiat is felt to be a sort of modifier of the first noun, which is clearly different from what happens in the case of N-of-N constructions, where the head is the second noun, and the first noun is predicated of the second noun.

Our speculation is that starting from the expression băiat de băiat, the pattern N-de-N came to be in fashion and grew productive, thus yielding expressions such as fată de fată (‘girl of girl’), mobil de mobil (‘mobile of mobile’) –more or less related to the world of uneducated, but wealthy people, whose life ideals are very down-to-earth. The pattern can now be used with any noun whatsoever.

(c) On a third hypothesis, the structure can be linked to the Genitive (Dumitrescu (2010)). It may be related to an older pattern that also involves the repetition of a noun, the second occurrence of which is genitive plural.

Traditionally associated to fairytales and archaic, popular language (23a), this pattern is

surprisingly productive in contemporary colloquial Romanian (23b):

(23) a. Şi deodată s-arăta / Păunaşul codrilor, / Voinicul voinicilor.

‘And suddenly CL-appears/ Peacock-the forest-GEN,/ Sturdy-the sturdy(N)-pl- GEN’

b. Ţăranul ţăranilor, oierul oierilor, prostu’ proştilor,

‘Peasant peasants-GEN, shepherd shepherds-GEN, fool-the fools-GEN, analfabetul, cine poate fi? Gigi, cel mai prost om din lume!

‘illiterate-the, who can be? Gigi, the more (most) stupid person in world!’

The two constructions have the same meaning: a superlative reading. “In Beyssade and Dobrovie-Sorin’s (2005) terms, the latter illustrates the intensional mode of predication, whereas the genitival construction is part of the extensional mode. If in the case of genitival superlatives like voinicul voinicilor the class reading obtains from the plural form of the second noun, in N

de N

such a reading requires the presence of the silent noun TYPE”

(Dumitrescu (2010)).

(d) On a fourth hypothesis, we can relate it to an N-N construction:

(24) El e băiat BĂIAT.

‘He is boy BOY.’

(25) E iubire IUBIRE între ei doi, nu glumă.

‘Is love LOVE between them two, no kidding.’

It can be argued that the preposition is inserted for reasons of case.

3.2. Back to Semantics and Syntax

Now the question which we would like to answer is whether the expressions under examination can be subsumed under the name of binominal constructions (i.e. qualitative pseudopartitives) or not. Our hypothesis is that they are not.

8

. Vidra, popular poem. http://ro.wikisource.org/

9

. http://www.sport.ro/europa-league/video-stoichita-antrenor-la-steaua-steaua-nu-se-refuza-bergodi-demis-

in-direct-la-sport-ro.html/pagina-20/

There are many arguments in favour of our hypothesis (semantic, syntactic, phonological, crosslinguistic).

1. Semantic arguments

(i) A first argument is related to the types of nouns that can occupy the positions in N

or N

. As already lengthily argued in section 2, in binominal constructions, the nouns which belong to N

are referential, while the nouns which belong to N

are evaluative (of N

).

Thus, if in the case of the nouns belonging to N

, any noun can more or less serve as a means of referring to an entity, the nouns belonging to N

represent a limited class. They have to be nouns which are scalar. Vişan (2003) proposes a test for such nouns, namely, their occurrence with aşa/ asemenea:

(26) N-am văzut un asemenea dobitoc/ prost not-have seen a such jackass/ fool

‘I haven’t seen a greater/ such a jackass/ fool…’

However, what we will claim is that this test is not that reliable actually, because, in fact, aşa/ asemenea can occur with any type of noun, not just with scalar nouns, and, moreover, it cannot be said that the nouns occurring in this context are coerced into a scalar interpretation, because this is really not the case. We can very well say something like:

(27) N-am văzut un asemenea iepure/ scaun/ dulap not-have seen a such rabbit/ chair/ wardrobe

and this certainly does not mean that we have turned iepure into a scalar noun. What we mean is that we have not seen such a lovely/ strange/ fluffy a.o. bunny. The context solves the mystery. Asemenea actually modifies an adjectives which is part of the shared knowledge of the speaker and the interlocutor. N-am văzut un asemenea dulap can easily be paraphrased as “Nu am văzut un dulap aşa de…”. And, in this case, no shift from non- scalar reading to scalar reading is at work.

In the case of expressions such as băiat de băiat, fată de fată, mobil de mobil, sentiment de

sentiment a.o., we can easily see that the nouns occurring in N

, N

can basically be any

type of noun whatsoever. There is no restriction as to the class of nouns that can occur in

N

, as there is in the case of binominals (where they have to be evaluative). If the set of N

nouns is not closed (nouns denoting concrete objects can very well be used, unlike in the case of binominals), this suggests that our construction is different.

(ii) A second argument is related to the interpretation of these expressions. In the case of binominals, we basically have the following interpretation: N

is referential, and N

is evaluative and it is predicated of N

. Following Beyssade and Dobrovie-Sorin (2005), we can argue that we are basically dealing with two modes of predication. By referring to a certain entity as being N

, we establish that the entity is a member of a set of entities (extensional mode of predication), while by ascribing the property N

to the entity referred to by N

, we localize a property in an entity(intensional mode of predication). From a semantic point of view, N

is predicative. However, this is not the way in which we interpret expressions such as this is not the way in which we interpret expressions such as băiat de băiat (‘boy of boy’), fată de fată (‘girl of girl’), trandafir de trandafir (‘rose of rose’), mobil de mobil (‘mobile of mobile’), masină de maşină (‘car of car’) a.o. In such expressions, the head is rather the first noun (N

), and the modifier of N

is N

:

(28) a. Toni şi-a cumpărat o maşină de maşină.

Toni-CLITIC (himself)-bought-a car-of-car

‘Toni bought himself a supercar.’

b. Toni şi-a cumpărat o minune de maşină.

Toni CLITIC (himself)-bought-a marvel-of car

‘Toni bought himself a marvel of a car.’

In (28 a), what the speaker is saying is that he bought a car which is worthy of the name of

“car”. The semantic head is the first noun, not the second (which acts as a modifier on the

first). In (28b.), what the speaker is saying is, once again, that he bought a car. However, in

this case, it is rather the second noun that is semantically selected by the verb a cumpăra

(‘to buy’), and not the first noun minunăţie (‘marvel’). The only difference is the position

of the head: in o maşină de maşină (‘a car of car’), it is the first noun, in o minune de

maşină (‘a marvel of car’), it is the second noun. In other words, o maşină de maşină (‘a

car of car’) means “o maşină care e cool (e MAŞINĂ)”, ‘a car that is cool (it is a CAR)’;

while o minune de maşină (‘a marvel of car’) means “o maşină care e o minune” (‘a car that is a marvel’).

2. Crosslinguistic argument

Another argument in favour of our hypothesis that such expressions are not in fact binominals is that there are no qualitative pseudopartitives in other languages that we are aware there are no qualitative pseudopartitives in other languages in which N

is the same as N

In other words, the opposite hypothesis that such expressions would be pseudopartitives is not at all supported by crosslinguistic data. As lengthily argued in the second section of the paper, English language does not allow expressions such as a boy of a boy, a beauty of a beauty, an idiot of an idiot, a.o. The reason for this is, as already explained, the avoidance of redundancies.

3. Phonological argument

Another argument is phonological: the different intonational contours and stresses which are ascribed to the constructions. The expressions o minune de maşină and o maşină de maşină are uttered in rather different ways, despite the fact that the words minune and maşină have the same number of syllables and the stress falls on the second syllable in both cases: /mi-nu-ne/, /ma-ʃi-nә/.

4. Syntactic arguments

Syntactic arguments can also be adduced to corroborate our hypothesis.

(i) In “Predicate Movement in Pseudopartitive Constructions” (1998), Norbert Corver enumerates some tests which any N-of-N construction should pass in order to count as a binominal construction.

(29) a. *Of a machine John bought a monster. (example from Corver (1998)) b. *De doctor am văzut un idiot.

‘*Of doctor have-I seen an idiot.’

c.*De băiat am văzut un băiat.

‘*Of boy have-I seen a boy.’

d. *Of extraordinary beauty he saw a girl.

Apparently, băiat de băiat behaves just like an N-of-N construction. However, this is not a reliable test. As we can clearly see in (29d), it is not the case that only qualitative pseudopartitives disallow the extraction of the of-phrase. Of extraordinary beauty cannot be extracted out of the phrase a girl of extraordinary beauty. In this case, a girl is the head, and the of-phrase is the modifier, unlike in the expression a monster of a machine. Hence, the fact that băiat de băiat (‘boy of boy’) successfully passes this test actually proves nothing whatsoever.

(ii) The verbs selects the second nominal in the DP.

(30) a. John drives [a monster of a truck].

b. #John rides [a monster of a truck].

c. John rides a monster. (examples from Corver (1998)) d. Ion a cunoscut o sărmaluţă de fată.

Ion has met a meat roll of girl.

e. #Ion a cunoscut o sărmăluţă.

Ion has met a meat roll.

f. Ion a cunoscut o fată de fată.

Ion has met a girl of girl.

g. Ion a cunoscut o fată.

Ion has met a girl.

In this case, the difference between the two constructions is pretty clear. If (30d) clearly does not entail (30e), (30f) entails (30g), which suggests that the two constructions have different heads (something already mentioned above, as a second argument in favour of our hypothesis.)

(iii) In N-of-N constructions adjectives enter into a modification relation with N

across N

.

(31) a. a nice bear of a fellow

b. a polite jewel of a child (examples from Corver (1998)) c. un simpatic băiat de băiat

‘a cute boy of boy’

d. o simpatică minunăţie de băiat

‘a cute marvel of boy’

The claim is that N

is “semantically transparent”. Hence, the adjective somehow modifies N

across N

. In other words, it is not the bear that is nice, but the [bear of a fellow], hence, the fellow. In a similar fashion, it is not the jewel that is polite, but the [jewel of a child], hence, the child. The N-of-N construction seems to have phrasal status. As we can clearly see in (31c), băiat de băiat can be preceded by an adjective, just like minunăţie de băiat.

However, it is not the case that the first noun in băiat de băiat is semantically transparent with respect to the adjective. On the contrary, in (31c), the adjective actually modifies the first noun (or, rather, the whole phrase).

(iv) In N-of-N constructions, N

is not a full-fledged DP.

(32) a. *that idiot of the/ that/ this/ my doctor b. *acel idiot de doctorul meu

‘*that idiot of doctor-the my’

c. *acel băiat de băiatul tău

‘*that boy of boy-the your’

(32c) is patently ungrammatical. But this does not make it an N-of-N construction, for it is not only in N-of-N constructions that N

is not a full-fledged DP. It suffices to think of phrases such as băiat de deputat (‘boy of deputy’) or fată de doctor (‘girl of doctor’).

Hence, this test again does not make băiat de băiat (‘boy of boy’) binominal in any way.

(v) N-of-N constructions allow recursivity:

(33) a. that asshole of an idiot of a doctor b. acea brută de catastrofă de avocat

‘that brute of disaster of lawyer’

c. acel prost de băiat de băiat

‘that stupid of boy of boy’

d. ?? băiat de băiat de băiat

‘boy of boy of boy’

e. *acel băiat de băiat de băiat

‘*that boy of boy of boy’

Indeed, băiat de băiat can in its turn be used as an N

in a binominal construction, just like catastrofă de avocat (‘disaster of lawyer’). But this tells us absolutely nothing about the internal structure of the expresssion băiat de băiat, whether it is binominal or not. What it tells us, in light of (iv) is that, given the fact that it can occupy the position of N

, it is not a full-fledged DP. Hence, it could be argued that băiat de băiat is rather an NP, it is predicational, not argumental. This is an important remark which we have to retain.

Interestingly, băiat de băiat cannot be used in a binominal in which N

is băiat (33e.). This makes perfect sense considering that the nouns occupying the second position in binominals have to be referential, whereas băiat de băiat is predicational. Equally interesting is the fact that (49d) is not entirely ungrammatical, if the first băiat is interpreted as the head, which is further modified by băiat de băiat.

(vi) In an N-of-N construction, the second noun cannot be removed out of the of-phrase.

(34) a. * a problem which this is a hell of b. *un kiwi care acesta e o minune de

‘* a kiwi that this is a marvel of’

c. *băiat care acesta e băiat de

‘*boy that this is boy of’

d. *calitate care aceasta e fată de

‘*quality that this is girl of’

This test, however, does not prove băiat de băiat binominal, since (vi) is true of

constructions which are not binominal, such as that in (34d). As we can see, our

expressions fail some significant tests that binominal constructions successfully pass ((ii)

and (iii)). As for the tests which they successfully pass, they do not prove them binominal

in any way, because other types of constructions pass them as well. Anyhow, the fact that

there are some tests which they fail is sufficient evidence for their non-binominal status,

which is exactly what we wanted to show. Moreover, the semantic and phonological tests

seem to be the most relevant.

3.3. Some Food for Thought

In the literature, qualitative pseudopartitives have generally been analysed (cf. Den Dikken (1998, 2006) and Corver (1998)) as involving predicate inversion. Unlike in the case of quantitative pseudopartitives, the attacks against a predicate inversion analysis of qualitative have not been so fierce. In the example below:

(35) a. A beauty of a woman invited Jim to a Magritte exhibition.

b. The woman is a beauty.

it can soundly be argued that predicate inversion is at work (the predicate comes before the referential noun, just like in inverted copular sentences).

Can the same thing be argued in the case of expressions such as băiat de băiat (‘boy of boy’), trandafir de trandafir (‘rose of rose’),a.o.? As already suggested above, it is not so.

Unlike un şmecher de băiat, deriving from (37b) (Dogaru (2009)), băiat de băiat does not derive from a sentence like (36b):

(36) a. băiat de băiat ‘boy of boy’

b. Băiatul e un băiat.

‘Boy-the is a boy.’

(37) a. Toni e un şmecher de băiat.

‘Toni is a cunning of boy.’

b. Băiatul e un şmecher.

‘Boy-the is a cunning.’

c. Băiatul e şmecher.

‘Boy-the is cunning.’

d. şmecherul de băiat

‘cunning-the of boy’

10

. For arguments in favour of a predicate inversion analysis, see den Dikken, “Predicate Inversion in the DP”

(1998), where he draws on the analogy between nominal and clausal structures to suggest that the predicate

moves across the referential noun within the nominal domain.

But can the same thing be said about un băiat de băiat (‘a boy of boy’)?

(38) a. Băiatul e un băiat.

‘Boy-the is a boy.’

b. Băiatul e băiat.

‘Boy-the is boy.’

c. un băiat de băiat a boy of boy (39) a. Toni e băiat de băiat.

‘Toni is boy of boy.’

b. ??Toni e un băiat de băiat.

‘Toni is a boy of boy.’

c. ???Băiatul de băiat a venit să mă vadă.

‘Boy-the of boy has come Conj-SUBJ me see’

d. Şmecherul de băiat a venit să mă vadă.

‘Maverick-the of boy has come conj- SUBJ me see.’

e. Am văzut un băiat de băiat.

‘Have-I seen a boy of boy.’

f. Am văzut un şmecher de băiat.

‘Have-I seen a maverick of boy.’

The first noun in our expressions is felt to be predicational. This is why adding the indefinite article in front of it is felt as rather odd (39b). Moreover, anaphorical use again sounds strange (39c). This is in stark contrast with N-of-N constructions (39d), (39f). This clearly points to a rather different status of N

in expressions such as băiat de băiat, an expression which is different both from un băiat de băiat, and from un idiot de băiat.

We would suggest the status of NP, a status which has already been hinted at when discussing the test (v)-Corver (1998). In addition, we notice that nothing can intervene between the two nouns:

(40) a. */?? acea cireaşă splendidă de cireaşă

‘that cherry splendid of cherry’

b. acea splendidă cireaşă de cireaşă

‘that splendid cherry of cherry’

c. acea supercireaşă splendidă

‘that megacherry splendid’

(41) *acea fată şmecheră de fată

‘that girl cunning of girl’

(42) acel idiot scârbos de politician

‘that idiot groce of politician’

After examining the data, we can thus remark that in expressions such as băiat de băiat (‘boy of boy’), fată de fată (‘girl of girl’), N

and N

have the following properties: (i) N

, the head, behaves like a predicate, like an NP, (ii) N

modifies N

and has to be adjacent to it; again, it is not a full-fledged DP. The modifier in băiat de băiat (‘boy of boy’) can no longer be modified, whereas the modifier in un idiot de băiat (‘an idiot of boy’) can be further modified. Therefore, different analyses should be proposed for the two constructions.

4. (Un) Băiat de Băiat as a Kind-Final Construction

We will make two claims: a) that (un) băiat de băiat (‘(a) boy of boy’) is a KIND construction; and b) that (un) băiat de băiat (‘(a) boy of boy’) is like “(a) boy of this type”

(KIND-FINAL), while un idiot de băiat (‘an idiot of boy’, “an idiot of a boy”) is like ‘a type of boy’ (KIND-INITIAL).

Several arguments can be adduced in favour of these claims. A first argument is represented by KIND PARAPHRASES:

(43) a. ‘maşină de maşină’ (‘car of car’) is like ‘car of (this) type’

while

b. ‘o frumuseţe de maşină’ (‘a beauty of car’) is like ‘a type of car’

A second argument is represented by EXTRACTION PHENOMENA, namely, these constructions behave like kind constructions with respect to extraction phenomena:

(44) I have bought a rose of this species.

(45) *Of this species I have bought a rose.

(46) Am cumpărat un trandafir de trandafir.

Have-I bought a rose of rose.

(47) *De trandafir, am cumpărat un trandafir.

Of rose, have-I bought a rose.

(48) *This species I have bought a rose of.

(49) *Trandafir, am cumpărat un trandafir de.

Rose, have-I bought a rose of

Thirdly, there is the phenomenon of KIND ANTI-ANAPHORA (Zamparelli (1998)), i.e.

the kind construction has particular anaphoric properties (or rather, does not have): the definite article has no anaphoric uses in kind constructions (kind-initial and kind-final).

(50) ?The tiger of that kind entered the room.

(51) The kind(s) of dog(s) *(we just mentioned) are/is quite popular.

In the same way as in (50), in Romanian, we have:

(52) ??Băiatul de băiat a intrat în cameră.

‘Boy-the of boy has entered in room.’

“The cool boy entered the room.”

A kind nominal with a simple definite article cannot be used to refer back to a previously introduced discourse referent, even if this is a kind

.

The article is in fact acceptable to the extent it is not used anaphorically:

(53) The types of contemporary poems are increasingly similar. (kind-initial)

(54) “The landlords of the traditional type had been supplemented by London based land- holding companies. (kind-final)

(55) Băiatul de băiat e un specimen des întȃlnit în ziua de azi.

‘Boy-the of boy is a specimen often encountered in day of today.’

“The cool boy is a specimen often encountered nowadays.’

Here, the definite article is not used anaphorically. Given the fact that both kind-initial and kind-final constructions have the property of KIND ANTI-ANAPHORA, the behaviour of băiat de băiat (‘boy of boy’) with respect to this test does not tell us whether the construction is kind-initial or kind-final.

11

. We have the following situations:

(i) kind-initial:

“Pink Delight”

and “Waverly” roses

were bred in England by Mr. Pinkerton. {*The / The two /These / His} kinds of roses

are quite popular nowaday in Scotland.

Trandafirii ‘Pink Delight’ şi ‘Waverly’ au fost cultivaţi în Anglia de Dl. Pinkerton. {*Tipurile/ /Cele două tipuri/ Aceste tipuri/ tipurile lui} de trandafiri sunt foarte populare azi în Scoţia.

(ii) kind-final:

The Greyhound

is common in England, although a dog of [{*the / this }kind]

always suffers in small spaces.

Greyhound-ul este des-înt ȃ lnit în Anglia, deşi un c ȃ ine de {*tipul/ tipul acesta} mereu suferă în spaţii

înguste.

5. A Syntactic Analysis of the Băiat de Băiat Construction

Starting from the above, we would like to propose a syntactic analysis of the băiat de băiat (‘boy of boy’), by putting together the Split D Hypothesis (Zamparelli (2000)) with Kayne’s (1994) antisymmetric representations of syntactic objects.

5.1. The Split D Hypothesis

In Layers in the Determiner Phrase (2000), Zamparelli aims at finding a common ground between:

(a) the idea that there should be a strict mapping between syntactic-semantic categories (Montague (1970, 1973));

(b)the fact that different NPs have different types of denotations ([John] (in John smiled) has the semantic type <e>, [a person] (in Mary is a person) has the semantic type <e, t>), [every dog] (in [Every dog] barked) has the semantic type <<e, t> t>)

The solution he proposes for explaining the predicative/ argumental uses of NPs is the SplitD Hypothesis. On this hypothesis, the interpretation of nominals does not depend on the position in the sentence in which they are interpreted (Kratzer (1989), Diesing (1992)), but on the position within the DP (Heim (1982), Reinhart (1987)).

The proposal has as a starting intuition the idea that the topmost part of the NP (determiners and quantifiers) includes 2 maximal projections (‘the determiner system’):

(a) the highest maximal projection, consisting of ‘strong’ determiners: PNs, personal pronouns (Milsark 1974), quantifiers

(b) the intermediate projection, consisting of ‘weak’ determiners: those determiners that can appear in predicate position.

The DP is thus split into three parts: SDP (Strong Determiner Phrase), PDP (Predicate

Determiner Phrase), KIP (Kind Phrase). Evidence for this tripartition comes from Italian,

which has a different pronoun for each layer (lo+ AGR for SDP, lo –AGR for PDP, ne for

KIP). PDP is the layer of weak determiners, by which we understand those determiners that

can occur normally in existential sentences: a, sm

, one, two, three, many, no. SDP is the

layer of strong determiners, by which we mean those determiners which cannot occur in

existential sentences: every, each, the, all, most, both, neither:

(56) [

(

Spec [

[strong det] [

Spec [

[weak det] [

Spec [

[of][

[N]]]]]]]]

5.2. What Syntactic Representation Can We Provide for the Construction Băiat de Băiat in a Split-DP Framework?

Taking the above into consideration, we try to provide a syntactic representation for băiat de băiat (‘boy of boy’) starting from the provided by Zamparelli (2000):

(57) [

Spec [

SD [

Spec [

PD

[

[

[

of] [

[

book] [

every kind]]]]]]]]

In the representation above, either every kind moves to [Spec, KIP], and then to [Spec, SDP], yielding every kind of book, or book moves to [Spec, KIP], yielding a book of every kind.

However, there are several problems with this analysis. The most important problem is that is NOT antisymmetrical: at the NP

level, NPj is made up of KIND and NP

(kind- initial constructions), and we would like an analysis that observes antisymmetry.

A second problem is that agreement facts are left unsolved, namely, the Agreement Generalization: *[[Arg]

[Pred]

]

(58) ‘this kind of tiger’

‘this kind of tigers’

‘*these kinds of tiger’

‘these kinds of tigers’

Our analysis is inspired from Zamparelli (1998), who proposes an RP analysis of partitives and possessives. RP is a Residue Phrase, i.e. a syntactic projection in charge of expressing the residue operation. In English the head of RP is realized as of. The two nouns can be

12

In sentences, we can have singular collective predicates (e.g. couple).

accommodated in the specifier and complement of this projection, where they are interpreted by the rule:

(59) [[RP ]]= Re ([[[Spec,RP]]], [[[Complement,RP] ]])

In turn, RP is embedded under a PDP, the site of numerals, and an SDP, to host external determiners (as in Every one of the boys). The phrase in [Spec,RP] is a KIP (cf. also Kayne (1994)), while the one in the complement is a full SDP:

(60) [

D [

two [

[

good friends] [

[

of] [

John [

[’s ] [

four [

good friends]]]]]]]]

As for băiat de băiat (‘boy of KIND boy’), the analysis we propose is:

(61) [

[

[

[

[

băiat [

[

de

[

TIP [

[KI] [

băiat (BĂIAT)]]]]]]]]]]

In the representation above, we make use of the silent noun TYPE. In our choice, we follow Dumitrescu (2010), who argues: “this position of TYPE is actually the one proposed in van Riemsdijk (2005) for Dutch N de N, but which seems inappropriate because it imposes the semantic head status on N

, therefore N

TYPE de N

was chosen as the right order in binominals. In contrast, in N

de N

the first noun is the head, so TYPE may be assumed to follow the preposition in this construction.” (Dumitrescu (2010))

In our analysis, both the Spec and the complement of R are KIP, and the RP can be embedded further on into a PDP, thus accounting for distributional facts (Am văzut un băiat de băiat, i.e., ‘I saw a boy of boy’).

Moreover, we would like to find a place for NumP in the structure. Dogaru (2007) argues that bare predicates are actually not bare, they are NumPs. A very important fact is that we have number agreement between the two nouns in băiat de băiat:

(62) băiat de băiat ‘boy of boy’

băieţi de băieţi

‘boys of boys’

*băiat de băieţi

‘boy of boys’

*băieţi de băiat

‘boys of boy’

We will propose placing NumP between KIP and NP:

(63) [

[

PD [

[

băiati [

[

de] [

TIP [

KI [

[

Num [

[

[

băiat (BĂIAT)]]]]]]]]]]]

This analysis presents many advantages, such as the fact that (i) it is antisymmetric, and the fact that (ii) it accounts for the fact that nothing can intervene between the two nouns:

(64) a. */?? acea cireaşă splendidă de cireaşă

‘that cherry splendid of cherry’

b. acea splendidă cireaşă de cireaşă

‘that splendid cherry of cherry’

c. acea supercireaşă splendidă

‘that megacherry splendid’

(65) *acea fată şmecheră de fată

‘that girl cunning of girl’

(66) acel idiot scârbos de politician

‘that idiot groce of politician’

13

. However, the question arises whether, considering that nouns are introduced as kinds (Zamparelli (2000)),

it is licit to place NumP between KIP and NP. Does it not break the unity KIP, NP?

6. What About the Other Languages? Do They Have a Construction of the băiat de băiat?

6.1. Some empirical facts

We would like to suggest that, although the băiat de băiat construction (‘boy of boy’) seems to be found in Romanian only, constructions with the same meaning and a similar form, or constructions with the same meaning but a different form exist in other languages as well.

In English, for example, we do not have this construction, but, instead, we have:

(a) the Saxon Genitive: a man’s man, a doctor’s doctor, a filmmaker’s film maker;

(b) the partitive: king of kings (animate, human), sun of suns (inanimate).

Apart from this, we encounter situations in which the same noun is repeated twice:. The rhetorical device is called epizeuxis, and it appears in exclamative/ deictic contexts: "The horror, the horror" (Kurtz in Heart of Darkness), "Simplicity, simplicity, simplicity!"

(Henry David Thoreau in Walden). The same thing can occur with adjectives: “Alone, alone, all all alone, /Alone on a wide, wide sea". (Samuel Coleridge in The Rime of the Ancient Mariner). However, in such cases, a comma is used. Moreover, if we have a DP, the whole DP is repeated: “the horror, the horror”, which is not the case in the construction băiat de băiat, or even băiat BĂIAT:

(67) a. *He is a boy BOY.

b. *He is a man MAN.

A possible explanation for this is that, in English, when two nouns are adjacent, the predicative/ attributive noun must appear before the noun that is the head.

However, we see that we do not even have:

(68) a. ??He is a BOY boy.

b. ???He is a MAN man.,

they are constructions that should be possible in principle (we have boyfriend, girlfriend, fisherman). We would like to argue that these are not ruled out by syntax, but by semantics/

pragmatics (in other words, what meaning could we assign to BOY boy, MAN man, what would they mean?)

If the noun is [+animate], [+human], then two constructions are possible (Saxon Genitive, partitive), whereas if the noun is [–animate], only the partitive is possible. (this is in accordance with the Saxon Genitive rules in English).

In Italian, we find two types of constructions:

(a) made of two nouns: uomo UOMO (??Lui è un uomo UOMO);

(b) the partitive: il re dei rei, il libro dei libri There is no Saxon Genitive construction in Italian.

In French, the partitive is present: la femme des femmes, le livre des livres, just as in Spanish: el libro de los libros.

In Brasilian Portuguese, we encounter:

(a) two nouns: um macho macho, (?) um homem homem;

(b) the partitive: o livro dos livros

In Chinese, we encounter a noun PRT noun construction: [jiu zhong zhi jiu] (‘wine middle PRT wine’), [mei-nii zhong de mei-nii] (‘pretty.lady middle PRT pretty lady’).

“Middle PRT” is the equivalent of the preposition of. The order is not “A of B” (as in ‘the wine of wines”), but “B middle PRT A”.

6.2. What is the syntactic structure for a man’s man?

In what follows, we would like to propose a syntactic structure both for the Saxon Genitive construction, and for the partitive construction, and then compare it to our construction.

Following Kayne (1994):

(69) [

[

definite] [

John [

[

–s][two pictures]]]]

what we have is:

(70) [

[

definite] [

man [

[

–s] man ]]

Adopting Zamparelli (2000)’s Split DP Hypothesis, we can further refine the above representation into:

(71) [

[

man] [

[

‘s] [

[

[

[

man]]]]]]

The structure above can preceded by PredP, or SDP (a man’s man, the man’s man).

Therefore, the Saxon Genitive construction (a man’s man) can be accounted for if we resort to a PossP (Possessive Phrase), while the băiat de băiat (‘boy of boy’) construction in Romanian can be accounted for by resorting to an RP (a Residue Phrase). The first construction includes a member in a class, while the second ascribes it a property.

6.3. What Is the Syntactic Structure for king of kings, sun of suns?

As for the partitive construction, we propose the following representation:

(72) [

[

[PD] [

[

king] [

[

of] [

[

[KI] [

[

[Num] [

kings]]]]]]]]]

Residue Phrase= Kayne’s Determiner Phrase

7. Conclusion

In conclusion, we would like to suggest that the băiat de băiat construction does not count as qualitative pseudopartitive (neither from a semantic, nor from a syntactic point of view).

The same noun is introduced in the structure twice, and the different meanings that the

nouns acquire in the structure are the result of the syntactic configuration in which they are

introduced. By combining Kayne (1994, 2005) and Zamparelli (2000), we have proposed a syntactic representation which manages to account for the ‘type’ reading of the second noun, and, hence, set them apart from qualitative pseudopartitives.

References

Brasoveanu, A. 2007. “Monotonicity in Pseudo-Partitives as a Consequence of Polysemy:

Evidence from Romanian”. Invited Presentation.

Corver, N. 1998. “Predicate Movement in Pseudopartitive Constructions”. In Alexiadou, A. and C. Wilder (eds.) Possessors, Predicates and Movement in the Determiner Phrase. 215-257. Amsterdam, John Benjamins.

den Dikken, M. 1998. “Predicate Invesion in DP”. In Alexiadou, A. and C. Wilder (eds.) Possessors, Predicates and Movement in the Determiner Phrase. 176-214.

Amsterdam, John Benjamins.

den Dikken, M. 2006. “Small Clauses and Copular Sentences”. In den Dikken, M. (ed.) Relators and Linkers: The Syntax of Predication, Predicate Inversion, and Copulas.

58-81. Cambridge, The MIT Press.

den Dikken, M. and P. Singhapreecha. 2004. “Complex Noun Phrases and Linkers”. Syntax 7.1:1-54.

Dogaru, M. 2009. MA Linguistics Course: “The Syntax and Interpretation of the DP”.

Dogaru, M. 2007. Chapter 4. “Interpretive Bare Nouns: Predicative Nominals”. In Phd Thesis. “The Category of Number. Its Relevance for the Syntax and the Semantic Typology of the Nominal Group”. Coord. Alexandra Cornilescu. 230-283. Bucharest, Editura Universitäţii din Bucureşti.

Dumitrescu, I. 2010. “Silent Nouns Have a Say in Romanian”. MA dissertation, University of Bucharest.

Kayne, R. 2005. “Silent Years, Silent Hours”. In Kayne, R. Movement and Silence. 241- 261(21). Oxford Scholarship Online Monographs.

Kayne, R. 1994. The Antisymmetry of Syntax. Cambridge, The MIT Press.

Mateescu, D. 2009. MA Linguistics Course: “Extrametrical Phonology”.

Matushansky, O. 2002. “A Beauty of a Construction”. 21

West Coast Conference on Formal Linguistics. UCSC, April 5-7, 2002. Santa-Cruz.

Schwarzschild, R. 2006. “The Role of Dimensions in the Syntax of NPs”. Syntax 9.1:67- 110.

Stickney, H. 2004. “The Pseudo-partitive, Extraposition and Predication within the DP”.

Presentation at the ECO5 Syntax Workshop. UMD.

Vişan, R. 2003. “Characterizing N de N Qualitative Constructions in Romanian”.

Bucharest Working Papers in Linguistics. Syntax. Semantics, vol. V. 137-148.

University of Bucharest Review.

Zamparelli, R. 1998. “A Theory of Kinds, Partitives, and OF/Z”. In Alexiadou, A. and C.

Wilder (eds.) Possessors, Predicates and Movement in the Determiner Phrase.

Amsterdam, John Benjamins. p:\\ftp.cogsci.ed.ac.uk\pub\roberto\ling\parkind.A4.pdf Zamparelli, R. 2000. Layers in the Determiner Phrase. New York, Garland Publishing

Edition.